Photograph: Matthew Korfhage
Two subtle ways agents can implicitly negatively affect the benchmark results but wouldn’t be considered cheating/gaming it are a) implementing a form of caching so the benchmark tests are not independent and b) launching benchmarks in parallel on the same system. I eventually added AGENTS.md rules to ideally prevent both. ↩︎
，这一点在新收录的资料中也有详细论述
Technical advice for open source maintainers
But in the last week, debates about how to make these decisions, as well as who is responsible for them, have broken into the mainstream. It comes after an Austrian man was convicted of gross negligent manslaughter over his girlfriend's death from hypothermia while climbing Austria's highest summit, Grossglockner, in January last year.
최강주 기자 [email protected]